I'm reading Lemma 7.2.1 of Ambrosio Gradient Flows in Metric Spaces 2nd edition.
On the top of page 159 we have
\begin{align*} W_p(\mu_0,\mu_1)\leq& \Big( \int_{X\times X} \|x_1-x_3\|^p d\mu(x_1,x_3) \Big)^{1/p} \\ \leq& ...? \\ =&\Big(\int_{X\times X} \|x_1-x_2\|^p d\gamma(x_1,x_2) \Big)^{1/p}+\Big(\int_{X\times X} \|x_2-x_3\|^p d\eta(x_2,x_3) \Big)^{1/p} \\ =& W_p(\mu_0,\mu_t)+W_p(\mu_t,\mu_1) \\ =& tW_p(\mu_0,\mu_1)+(1-t)W_p(\mu_0,\mu_1)=W_p(\mu_0,\mu_1). \end{align*}
But I cant fill in the blanks in the line of argument, to be honest I don't understand the notation $\|x_1-x_3\|_{L^p(\mu,X)}$ I would read this as the Lp norm of function $x_1$ but obviously it doesn't mean this. Can someone explain the reasoning?
This is a shortcut for $\|f_1-f_3\|_{L^p(\mu,X)}$ with $f_1(x_1,x_3)=x_1$ and $f_3(x_1,x_3)=x_3$.